summree
Claude Video is Here… Automate Anything
Claude
Jack Roberts

Claude Video is Here… Automate Anything

⏱ 18 min video · 3 min read7 May 2026Worth watching
TL;DR
Jack Roberts demonstrates how to connect Claude Code and the Hixfield MCP/CLI to create a unified AI content operating system that generates images, videos, and websites from a single environment. The core value proposition is eliminating the need to juggle dozens of separate AI tools by routing everything through one workflow.
Key points
1
Hixfield MCP/CLI connects Claude Code or Codex to 50+ generation models (Nano Banana, Flux, Runway Gen 4, Kling AI, Sora V3, etc.) through one integration, so switching models requires no re-engineering
2
Install the Hixfield CLI (recommended over MCP or SDK) so any environment — Codex, Claude Code — draws on the same connection and all models remain accessible
3
Claude Code builds the website, Hixfield generates 4x5 MacBook images in 4 versions using Nano Banana Pro (auto-selected for dimension compatibility), then video animation is integrated as a scroll-triggered effect — all without leaving the Codex environment
4
Reusable skill system: after running any workflow once, prompt Claude to save it as a skill.md file so it auto-runs on future requests — building a compounding content library over time
5
Hixfield pricing starts at $15/month (annual) for 200 credits; Nano Banana 2 image at 2K costs 2 credits per image, a 5-second Kling video costs ~45 credits
Actionable insights
Install Hixfield via CLI (not MCP or SDK) — it lets every environment share one connection and exposes the widest model selection including batch image generation that single-model integrations like GPT Image 2 CLI cannot match
Use two GitHub design repos as context when prompting Claude for websites: a public design collective repo (includes Spotify system) and a power design principles repo — this dramatically improves output quality without manual design work
Build a character in Hixfield using 20+ personal photos, then save a skill.md that auto-references your character ID whenever you request a photo of yourself, enabling custom LinkedIn-style content generation on demand
After any successful workflow, immediately prompt: 'Turn this into a reusable skill' — Claude saves the steps as skill.md and the workflow becomes callable in future sessions without re-explaining it
After generating a website, publish in one step by prompting Claude Code to create a GitHub repo via GitHub CLI and deploy via Vercel CLI — no manual steps required
Notable quotes

I've seen Higsfield everywhere. I see Higsfield online now more than my own mom.

Run the workflow once, save as a skill.md, and then it will auto run the next time.

Worth watching?
Worth watching the full video?
Watch if you want to see the live build in Codex — the prompts, model selection logic, and scrolling animation integration are easier to follow visually, though the key steps are all captured here.
Topics
AI & TechClaude

Explore more summaries on these topics →

Saved you some time? The creator still deserves a like.

Watch on YouTube →
More like this

Want this for your own channels?

Add the channels you follow. Every new video summarised and in your inbox the moment it drops. From £4/month.

Try it free